Development of a Multidimensional Instrument of Person ...web.ba.ntu.edu.tw/professor/contents/aichia/ChuangShenJudge2016APIR.pdf · multidimensional (Edwards & Billsberry, 2010;

Development of a Multidimensional Instrument ofPerson–Environment Fit: The PerceivedPerson–Environment Fit Scale (PPEFS)

Aichia Chuang*National Taiwan University, Taiwan

Chi-Tai ShenNational Changhua University of Education, Taiwan

Timothy A. JudgeUniversity of Notre Dame, USA,

University College London, UK and

King Abdulaziz University, Saudi Arabia

This research identifies four challenges in the field of person–environmentfit (PE fit): the multidimensionality of PE fit, the integration of fit theories,the simultaneous effects of the multiple dimensions, and the function of thedimensions. To address those challenges, we develop a theory-driven and sys-tematically validated multidimensional instrument, the Perceived Person–Environment Fit Scale (PPEFS), consisting of four measures: the Person–JobFit Scale (PJFS), the Person–Organisation Fit Scale (POFS), the Person–Group Fit Scale (PGFS), and the Person–Supervisor Fit Scale (PSFS). Data arecollected from 532 employees and 122 managers for two independent studieswith multiple rater sources and multiple time points. A series of validationanalyses and hypothesis tests reveals that the PPEFS measures have goodpsychometric properties (i.e. reliability, convergent validity, discriminant valid-ity, and criterion-related validity) and exhibit incremental validity above and

* Address for correspondence: Aichia Chuang, Department of Business Administration, College ofManagement, National Taiwan University, RM 811, 85, Sec. 4, Roosevelt Road, Taipei, Taiwan 106.Email: [email protected]

This research was supported in part by a grant from the National Science Council in Taiwanto Aichia Chuang for contract number NSC 95-2416-H-002 -016 -MY3. An early version of thisarticle was presented at the Annual Meeting of the Academy of Management, Philadelphia,August 2007. We are grateful to the editor and two anonymous reviewers for their insightful andconstructive comments throughout the review process. We thank Huichen Hsu for her valuableassistance during the process of this project.

APPLIED PSYCHOLOGY: AN INTERNATIONAL REVIEW, 2014doi: 10.1111/apps.12036

© 2014 International Association of Applied Psychology.VC 2015 International Association of Applied Psychology.

APPLIED PSYCHOLOGY: AN INTERNATIONAL REVIEW, 2016, 65 (1), 66–98doi: 10.1111/apps.12036

beyond Cable and DeRue’s (2002) fit measures. Furthermore, the measures arereflected by a superordinate (vs. aggregate) construct of PE fit. Overall, the fourdifferent types of fit significantly predict in-role behavior, job satisfaction,intent to quit, and organisational citizenship behavior (OCB), each explainingthe greatest amount of variance in different outcomes. The PPEFS shouldprove useful in future research regarding PE fit.

INTRODUCTION

. . . recent advances in fit theory have recognized that the most rewarding experi-ences are those in which multiple types of fit exist simultaneously. That has ledto conceptual work exploring PE fit as a multidimensional construct . . .(Kristof-Brown & Guay, 2011, p. 13)

The match between individuals and the environment, or PE fit, has longbeen a research topic of interest to industrial and organisational psycholo-gists (Kristof-Brown, Zimmerman, & Johnson, 2005). Academicians arenot alone in their interest; PE fit has attracted the attention of recruiters,job seekers, and incumbent workers in the business world (Kristof-Brown,2000). Over the decades of PE fit research, four types of fit have emergedas the most studied phenomena (Kristof-Brown & Guay, 2011): person–job fit (PJ fit), person–organisation fit (PO fit), person–group fit (PGfit), and person–supervisor fit (PS fit). These dimensions of PE fit havecontributed to the literature on work attitudes, turnover, performance,job search, and managerial selection decisions (Kristof-Brown et al.,2005).

The concept of person–environment fit is grounded in the interactionisttheory of behavior. Early works such as Pervin (1968) rested upon theassumption that certain environments correspond to each individual, mostlymatching the characteristics of the individual’s personality, and that thiscorrespondence, in turn, results in higher performance, higher satisfaction,and less stress for the individual. Since Pervin, research using diverse repre-sentations of fit has proliferated in support of the validity of PE fit(Kristof-Brown et al., 2005). However, four important challenges requirefurther attention: the consideration of multiple dimensions of PE fit, theintegration of PE fit theories, the simultaneous assessment of the effects ofthe multiple dimensions, and the analysis of the function of the dimensions.Our study addresses these challenges by employing two independent studiesto develop a multidimensional scale of perceived PE fit: the PerceivedPerson–Environment Fit Scale (PPEFS). This scale is composed of fourmeasures: the Person–Job Fit Scale (PJFS), the Person–Organisation FitScale (POFS), the Person–Group Fit Scale (PGFS), and the Person–Supervisor Fit Scale (PSFS). The following sections will address each of thefour challenges.

2 CHUANG ET AL.

© 2014 International Association of Applied Psychology.

MULTIDIMENSIONAL INSTRUMENT OF PERSON–ENVIRONMENT FIT 67

VC 2015 International Association of Applied Psychology.

MULTIDIMENSIONALITY OF PE FIT

The first challenge pertains to the dimensionality of PE fit. Though PE fitliterature has been an active body of research for decades, the field was oncecharacterised as “elusive”, indicating that researchers in the field remainedgenerally unclear about the construct of fit. This criticism continued untilquite recently, when an increasing number of studies theorised PE fit asmultidimensional (Edwards & Billsberry, 2010; Jansen & Kristof-Brown,2006; Wheeler, Buckley, Halbesleben, Brouer, & Ferris, 2005). The mainargument of this line of research is that studying fit from only a singledimension is inconsistent with how individuals experience fit because peopleare simultaneously nested in multiple aspects of an environment. Recentadvances in fit research have described this integrative view as the “nested” or“holistic” view (Jansen & Kristof-Brown, 2006; Kristof-Brown & Guay,2011). Ostroff, Shin, and Kinicki (2005) asserted that focusing on only one ora few types of fit generates a limited picture of the effects of fit becausedifferent types of fit have been revealed to have varying effects on employeeattitudes and behavior.

In addition to focusing on the multidimensionality of PE fit (e.g. PJ, PO,PG, and PS), researchers should also consider the multiple content dimen-sions (e.g. values, goals, personality, and interests) of each individual dimen-sion of PE fit. Edwards and Cooper (1990) argued that many researchers hadcovered only a very limited number of content dimensions. Such minimalcoverage creates problems including content validity issues and incompletedeterminants of criteria. In a review regarding the conceptualisation andmeasurement of perceived PO fit, Piasentin and Chapman (2006) called forfuture research to focus on clearly measuring different characteristics of fit,such as values and goals. To address this issue, we have developed a new scalethat addresses multiple types of PE fit and multiple content dimensions foreach type of fit.

A second challenge in the current literature is the integration of PE fittheories. For many years, researchers have studied PE fit on the basis of asingle dimension. This approach has allowed for a thorough and in-depthinvestigation of each dimension of fit, but this tendency has obscured ourunderstanding of the simultaneous theoretical effects of fit on major out-comes. As early as Kristof (1996), the literature was calling for integration ofmultiple theories of fit. As noted in our opening quote, recent developmentsin PE fit research have led to theorisations of fit based on multiple theories.The integration of different fit theories would allow researchers to paint aricher portrait of PE fit phenomena and investigate the unique effects of eachtheory on these phenomena (Kristof-Brown & Guay, 2011).

Most current efforts to integrate PE fit theories involve combinations oftwo or three dimensions of fit (Cable & DeRue, 2002; Greguras &



68 CHUANG ET AL.


Diefendorff, 2009; Kristof-Brown, Jansen, & Colbert, 2002; Lauver &Kristof-Brown, 2001; Wang, Zhan, McCune, & Truxillo, 2011). Otherstudies have incorporated four or more dimensions (Edwards & Billsberry,2010; Jansen & Kristof-Brown, 2006; Wheeler et al., 2005). Following thisline of research, our study develops a four-dimension PE fit scale groundedon the complementarity-based view (Muchinsky & Monahan, 1987), Hol-land’s theory on occupational interests (Holland, 1996), the need-fulfillmentparadigm (French & Kahn, 1962), the attraction-selection-attrition frame-work (Schneider, 1987), the similarity attraction paradigm (Byrne, 1971), andthe interpersonal attraction theory (Huston & Levinger, 1978).

A third challenge involves simultaneously assessing the contributions ofvarious types of PE fit to the theoretically related outcome constructs.Kristof-Brown and Guay (2011) have asserted that “multicollinearity isoften a concern when determining the unique impact of various types offit” (p. 37). This is confirmed by meta-analytic results showing moderate tohigh correlations between types of perceived fit (Kristof-Brown et al., 2005;Oh, Guay, Kim, Harold, Lee, Heo, & Shin, 2014). The current practice ofpulling scales of different fit types from separate sources may render thevarious scales indistinguishable, increasing the likelihood of high correla-tions among the various types of fit. Given these conditions, this studyintends to develop a multidimensional PE fit instrument in which theincluded content dimensions are well thought out and empirically proven tobe distinct.

The final challenge found in the PE fit literature pertains to assessments ofthe function (e.g. superordinate vs. aggregate) of the multidimensional con-struct of PE fit. With an integrated multidimensional fit scale, researchers inthe field can further explore how different types of fit correspond to a higher-order construct of PE fit. Doing so would provide them with a more realisticview of the work experience. In general, multidimensional constructs can bedistinguished by the direction of the relationship between the construct andits dimensions (MacKenzie, Podsakoff, & Podsakoff, 2011). A construct isdescribed as superordinate when the relationships flow from the construct toits dimensions, in which case the construct is manifested by specific dimen-sions. In contrast, a construct is described as aggregate when the relation-ships flow from the dimensions to the construct, in which case the constructcombines specific dimensions into a general concept. Superordinate con-structs are widespread in organisational research on such topics as g-factor,personality, and general work values. Aggregate constructs are also commonin literature regarding job satisfaction, job performance, and job character-istics. However, results regarding the function of PE fit have been mixed. Assuch, our study attempts to advance this line of investigation by developinga theoretically driven multidimensional PE fit instrument and by testing itsfunction.

4 CHUANG ET AL.




THEORIES, FIT DIMENSIONS, AND HYPOTHESES

The current study develops the content dimensions of each type of fit on thebasis of corresponding theories and an extensive review of the literature onPE fit. This section addresses the dimensionality of each type of fit underexamination, and proposes hypotheses for the effects of the different types offit and for their functions. The outcome variables consist of employee in-rolebehavior, job satisfaction, intent to quit, and OCB. These variables aretheoretically related to person–environment fit concepts, commonly cited asoutcomes in Kristof-Brown et al.’s (2005) meta-analytical article, and denotea fair representation of attitude variables, performance outcomes, andturnover-intention indices.

Person–Job Fit

“Person–job fit” is broadly defined as an individual’s compatibility with aspecific job (Kristof, 1996). It is specifically defined by Edwards (1991) as thefit between the demands of a job and the abilities of an individual (demands–abilities fit or “DA fit”), or the needs of a person and the supplied attributesof a job (needs–supplies fit or “NS fit”). We argue that DA fit consists of thedimensions of KSAs (knowledge, skills, and abilities) and personality, andNS fit consists of the dimensions of interests and job characteristics.

In relation to DA fit, according to the complementarity-based view(Muchinsky & Monahan, 1987), one complements the characteristics of anenvironment when one’s ability matches the job requirement. The mostcommonly used content definition for DA fit is the “KSAOs” (Bretz, Rynes,& Gerhart, 1993; Kristof-Brown, 2000). The acronym stands for Knowledge,Skill, Ability, and Other characteristics. Bretz et al. examined interview tran-scripts and found that the most frequently mentioned determinant of PJ fitwas job-related coursework or experience. In addition, jobs may requireincumbents to possess a certain type of personality for better performance.Piasentin and Chapman (2006) postulated that personality characteristicsand work-related skills/abilities may be important for assessing complemen-tary fit perceptions. Past research has also used personality to measure per-ceived job fit (Lauver & Kristof-Brown, 2001).

With regard to NS fit, Holland’s theory of occupational interests (Holland,1996) could be used to include interests as a content dimension for NS fit.Holland’s theory involves six personality types that describe individuals’career interests and their environment: realistic, investigative, artistic, social,enterprising, and conventional. In general, interests congruence is contribu-tive to positive outcomes at work, such as satisfaction, retention, and accom-plishment. The need-fulfillment paradigm (French & Kahn, 1962) suggeststhat individuals compare their own needs (e.g. recognition and social involve-ment) with environmental supplies. It rests upon the proposition that people



70 CHUANG ET AL.


experience more positive work outcomes when their needs are fulfilled byenvironmental supplies. Kulik, Oldham, and Hackman (1987) proposed thatPE fit (referring specifically to PJ fit in their article) can be a good matchbetween certain job characteristics (skill variety, task identity, task signifi-cance, autonomy, and job feedback) and certain characteristics of the indi-vidual (knowledge and skills, growth needs, strength, and satisfaction withwork context).

In sum, based on the complementarity-based view, Holland’s (1996)theory on occupational interests, and the need-fulfillment paradigm, thepresent study contains four job fit dimensions: KSAs, personality, interests,and job characteristics. Empirical evidence has revealed that PJ fit is relatedto job/task performance (Li & Hung, 2010; Wang et al., 2011), job satisfac-tion (Cable & DeRue, 2002; Wang et al., 2011), turnover intention (Wanget al., 2011), turnover decision (Cable & DeRue, 2002), and OCB (Li &Hung, 2010). Thus, we propose the following.

Hypothesis 1: PJ fit is related to in-role behavior (Hypothesis 1a), job satisfaction(Hypothesis 1b), intent to quit (Hypothesis 1c), and OCB (Hypothesis 1d).

Person–Organisation Fit

“Person–organisation fit” is defined as congruence between an individual andhis or her organisation in terms of such dimensions as values and goals(Kristof, 1996). Theoretically, Schneider’s (1987) attraction-selection-attrition framework, upon which much PO fit research has drawn, states thatpeople are attracted to and selected by an organisation with which they sharevalues and attributes, a match which subsequently generates PO fit. When thematch no longer exists, people opt to leave the organisation. In fact, manyPO fit studies contained only the value dimension (e.g. Cable & DeRue,2002), while others specifically targeted goal congruence (Vancouver &Schmitt, 1991). However, PO fit dimensions comprising values and goalshave been used in previous research (Chuang & Sackett, 2005). Piasentin andChapman (2006) reviewed 46 empirical studies that measured PO fit percep-tions. Of these studies, 78 per cent included value congruence variables and20 per cent included goal congruence variables. Hence, the current studyproposes both values and goals as dimensions of perceived PO fit.

The attraction-selection-attrition framework in relation to PO fit has beenshown to be associated with employee job performance (Kim, Aryee, Loi, &Kim, 2013), job satisfaction (Cable & DeRue, 2002; Kim et al., 2013;McCulloch & Turban, 2007; Vancouver & Schmitt, 1991; Wang et al., 2011),turnover intention (Vancouver & Schmitt, 1991; Wang et al., 2011),employee retention (McCulloch & Turban, 2007), and citizenship behaviors(Cable & DeRue, 2002; Kim et al., 2013). Thus, we hypothesise the following.

6 CHUANG ET AL.




Hypothesis 2: PO fit is related to in-role behavior (Hypothesis 2a), job satisfaction(Hypothesis 2b), intent to quit (Hypothesis 2c), and OCB (Hypothesis 2d).

Person–Group Fit

“Person–group fit” has been defined as compatibility between individualsand their work group (Kristof, 1996). The similarity attraction paradigm(Byrne, 1971) may help clarify the phenomena of PG fit. This paradigmpredicts that a person is generally attracted to similar others in his or hersocial milieu. A large body of research has investigated how similarity inattitudes, personality, values, and goals can facilitate attraction. The reasonfor similarity-based attraction may be that similarity appears to help predicthow other people behave. Research has found that personality similarityhelps facilitate communications among employees and foster social integra-tion (Schaubroeck & Lam, 2002).

Most of the PG fit studies are quite specific about the characteristics beingexamined, focusing on such issues as person–group personality fit (Seong &Kristof-Brown, 2012), values fit (Seong & Kristof-Brown, 2012), goals fit(Kristof-Brown & Stevens, 2001), or work style fit (Kristof-Brown et al.,2002). Prior group composition studies have also shown that behavioral andattitudinal outcomes for groups and members can be traced back to thecomposition of group members in terms of values (Harrison, Price, Gavin, &Florey, 2002), goals (Shaw, 1981), personality (Harrison et al., 2002), workstyle (Riordan, 2000), and lifestyle (DiMarco, 1975). Hence, the currentstudy includes values, goals, and group member attributes (personality, workstyle, and lifestyle) as dimensions of person–group fit.

Empirical research has shown that PG fit is significantly related to indi-vidual performance (Kristof-Brown & Stevens, 2001), job satisfaction (Wanget al., 2011), turnover intention (Wang et al., 2011), and OCB (Seong &Kristof-Brown, 2012). Therefore, we postulate the following.

Hypothesis 3: PG fit is related to in-role behavior (Hypothesis 3a), job satisfaction(Hypothesis 3b), intent to quit (Hypothesis 3c), and OCB (Hypothesis 3d).

Person–Supervisor Fit

“Person–supervisor fit” denotes the match between an individual and his orher supervisor in a work environment, and is by far the most well-studieddyadic fit in a work setting (Kristof-Brown et al., 2005). Interpersonal attrac-tion theory (Huston & Levinger, 1978) explains that an individual is attractedto another individual on the basis of similar characteristics regarding lifegoals, personality, activity preferences, values, and so on. A subordinate anda supervisor who are attracted to each other on the basis of such similarity



72 CHUANG ET AL.


are said to “fit” each other. Interpersonal attraction has been an importantresearch topic in organisational psychology. Previous PS fit research hasexamined values (Hoffman, Bynum, Piccolo, & Sutton, 2011; Van Vianen,2000), personality (Schaubroeck & Lam, 2002), work style (Turban & Jones,1988), lifestyle (DiMarco, 1974), and leadership style (Chuang, Judge, &Liaw, 2012). Drawing on these findings, the current study has operationalisedthe content dimensions of PS fit to include values, personality, work style,lifestyle, and leadership style.

Research has found PS fit to be related to in-role performance (Huang &Iun, 2006), job satisfaction (Ostroff et al., 2005), turnover intention (Ostroffet al., 2005; Van Vianen, 2000), and OCB (Huang & Iun, 2006). Therefore,we posit the following.

Hypothesis 4: PS fit is related to in-role behavior (Hypothesis 4a), job satisfaction(Hypothesis 4b), intent to quit (Hypothesis 4c), and OCB (Hypothesis 4d).

Function (Superordinate vs. Aggregate) ofPE Fit Dimensions

Research has recently begun to explore the function of the relationshipsamong fit dimensions, both conceptually (Jansen & Kristof-Brown, 2006)and empirically (Edwards & Billsberry, 2010; Seong & Kristof-Brown, 2012;Seong, Kristof-Brown, Park, Hong, & Shin, in press). The implications fromthe explorations have been mixed. On one hand, PE fit dimensions couldform an aggregate model. Jansen and Kristof-Brown introduced the conceptof multidimensional PE fit to reflect the fact that individuals are faced with aholistic environment, which encompasses various aspects. They proposed acombined (aggregate) model where the overall PE fit is an algebraic combi-nation of multiple dimensions. Supporting this assertion, research has theo-rised that different types of fit are distinguishable and lead to differentoutcomes. For instance, Kristof-Brown (2000) found that recruiters are ableto distinguish between applicants’ PJ fit and PO fit, and those two types offit offer unique predictions of hiring recommendations. In Lauver andKristof-Brown (2001), employees’ perceptions of PJ and PO fit were found tobe distinct constructs, each of which has a unique impact on job satisfactionand intent to quit. Similarly, Cable and DeRue (2002) showed that employeescan differentiate between needs–supplies fit, demands–abilities fit, and PO fit,each of which has a different pattern in predicting outcome variables. Theforegoing conclusions have led the field to believe that the various types of fitare theoretically distinct, rendering an aggregate model of PE fit. Therefore,we propose the following.

Hypothesis 5a: PE fit is an aggregate multidimensional construct.

8 CHUANG ET AL.




On the other hand, empirical results based on tests of a multidimensionalconstruct of PE fit have provided evidence for alternative functional forms.Edwards and Billsberry (2010) tested Jansen and Kristof-Brown’s (2006)aggregate model against a competing model, in which multiple dimensions offit were proposed to be distinct, assuming no overarching sense of fit. Theauthors found that the aggregate model fit poorly, whereas the distinct modelfit well. Those authors did not, however, test a superordinate model. Addingto the complication, Seong and Kristof-Brown (2012) tested a multidimen-sional model of individual-level PG fit and found that PG fit was better suitedto a superordinate model than to an aggregate or distinct model. Similarly,Seong et al. (in press) found that group-level PG fit is a superordinate con-struct, manifested by subdimensions of supplementary and complementaryfit. These results suggest that fit dimensions may be driven by a higher-orderperception of overall fit. Hence, we propose a competing hypothesis.

Hypothesis 5b: PE fit is a superordinate multidimensional construct.

GENERAL METHOD

Overview and Participants

Two separate studies covering a total of 532 employees and 122 managerswere employed for the development of the Perceived Person–EnvironmentFit Scale (PPEFS). These samples were collected over multiple points in timeand represent a diverse cohort of respondents from various industries andorganisations. Specifically, Study 1 contained data from 328 employees and67 managers and was used to conduct all analyses described in the “AnalysisStrategies” section, below. Study 2 used a sample from the service industrythat consisted of 204 service-type employees and 55 managers. Study 2 servedto replicate the results of Study 1 regarding discriminant validity andcriterion-related validity.

Measures of the PPEFS

All items used in the PPEFS were newly developed for the purpose of thisresearch. The PJFS consists of DA fit (KSAs and personality) and NS fit(interests and job characteristics). The POFS consists of two dimensions:values and goals. We used the four general values (i.e. honesty, achievement,fairness, and helping others) which have been shown to be operative in theworkplace (Ravlin & Meglino, 1987) with the Comparative Emphasis Scale.For goals, we adopted Van Vianen’s (2000) concepts of Organisational GoalOrientation (i.e. reward, effort, and competition). For the PGFS, we devel-oped 10 items covering the content dimensions of values (Comparative



74 CHUANG ET AL.


Emphasis Scale), goals (Organisational Goal Orientation), and memberattributes. For the PSFS, we developed five items comprising the contentdimensions of values, personality, work style, lifestyle, and leadership style.The Appendix contains a list of the final items of the PPEFS.

Analysis Strategies

To ensure the psychometric properties of the PPEFS and to test the hypoth-eses, we conducted a series of analyses suggested in the scale developmentliterature (Hinkin, 1998; Judge, Erez, Bono, & Thoresen, 2003; Pierce,Gardner, Cummings, & Dunham, 1989). In the following, we elaborate oneach of the analyses performed.1

Stage 1: CFA and Reliability. We performed a CFA using LISREL 8.54to corroborate the structure of the model (Hinkin, 1998). Models with χ2/df

equal to 2 or 3, RMSEA less than .08, and CFI and IFI greater than .90 areacceptable (Bentler & Bonett, 1980). Based on the recommendation ofAnderson and Gerbing (1988), we formed alternative models and comparedthem with the hypothesised model by using chi-square difference tests. Weestimated reliability using Cronbach’s alpha. A newly developed scale withan alpha of .70 or higher is deemed acceptable (Nunnally, 1976).

Stage 2: Convergent, Discriminant, and Criterion-Related Validity. Ascale has convergent validity when it significantly correlates with anotherexisting scale of the same construct (Hinkin, 1998). For each fit type, we useda relevant existing fit scale. Discriminant validity is ensured when the scalehas a weak or negligible correlation with other theoretically unrelated meas-ures assessed by the same source (Hinkin, 1998). In this study, we correlatedthe fit measure with employee demographics (i.e. age and gender) because nocompelling theory implies that employees of different demographics willexperience different levels of fit (Liao & Chuang, 2004). Criterion-relatedvalidity is achieved when the scores of a scale correlate with theoreticallyrelated constructs or variables in the nomological net (Hinkin, 1998). For allthree validation methods, we calculated Pearson correlations.

Stage 3: Usefulness Analyses. Usefulness analyses (Judge et al., 2003)are used to determine whether a scale has incremental validity beyond exist-ing measures. The present study used a relevant existing fit measure and thePPEFS to predict multiple outcome criteria that are theoretically related toPE fit (i.e. in-role behavior, job satisfaction, intent to quit, and OCB). Using

1 From stage 1 to stage 3, we tested the four fit measures individually, and from stage 4 tostage 6, we tested the four fit measures simultaneously.

10 CHUANG ET AL.




hierarchical regressions, we conducted two procedures. For procedure 1, wefirst entered the existing fit measure to predict the criterion variables, andthen added the corresponding PPEFS measure to the regressions to investi-gate the change in multiple correlations. A significant ΔR2 indicates theincremental validity of—and, thus, the usefulness of—the PPEFS measure.We also performed the reverse situation (procedure 2), first entering thePPEFS measure and then entering the existing fit measure.

Stage 4: Relations among and Convergent/Discriminant Validity of the

PPEFS Measures. We investigated the relations among the individualPPEFS measures by computing bivariate correlations. In addition, we exam-ined the convergent and discriminant validity of our PPEFS measures, allsimultaneously present in one model, following procedures recommended byAnderson and Gerbing (1988). Specifically, we estimated a second-ordermeasurement model where the second order contained the four PPEFS meas-ures (i.e. PJFS, POFS, PGFS, and PSFS) and the first order included twosubscales of POFS and three subscales of PGFS. Convergent validity isassessed by determining “whether each indicator’s estimated pattern coeffi-cient on its posited underlying construct factor is significant (greater thantwice its standard error)” (Anderson & Gerbing, 1988, p. 416). Discriminantvalidity is assessed by determining “whether the confidence interval (± twostandard errors) around the correlation estimate between the two factorsincludes 1.0” (Anderson & Gerbing, 1988, p. 416). To determine whether oursecond-order model was better, we applied chi-square difference tests andcompared our model with two competing models: one with the first-ordersubscales removed (four-factor model) and the other with all items linkingdirectly to a general factor (one-factor model).

Stage 5: Contributions of the PPEFS Measures. To test Hypotheses1a–1d, 2a–2d, 3a–3d, and 4a–4d and to simultaneously test the effects of thePPEFS measures, we adopted Johnson’s (2000) relative weight analysis totake into consideration the intercorrelations among the PPEFS measures.This method can be used to estimate the relative weight of the predictors andis particularly useful when the predictors have non-zero intercorrelations.The “relative weight” of each predictor refers to the proportionate contribu-tion each predictor makes to R2, taking into account both the predictor’sunique effect and its effect when combined with other variables. We alsorescaled the weight for each predictor by dividing the relative weight of eachpredictor by the total R2 of the full model. In doing so, the sum of the relativeweights of the predictors was equal to R2, improving the interpretabilityof the relative importance of the predictors. In this research, we used multi-ple regressions and regressed each outcome variable on the employees’



76 CHUANG ET AL.


demographic variables (i.e. age, gender, education, and tenure) and the fourmeasures of the PPEFS (cf. Oh et al., 2014).

Stage 6: Function (Superordinate vs. Aggregate) of the PPEFS Meas-

ures.2 To test Hypotheses 5a and 5b, we followed suggestions in the extantliterature regarding estimating the functions of multidimensional models(MacKenzie et al., 2011; Seong & Kristof-Brown, 2012; Tang & Sutarso,2013) and assessed an aggregate and a superordinate model (Figure 1) of thePPEFS measures. We then compared the fit indices of these two models andtested whether each indicator’s estimated coefficient on its posited underlyingconstruct factor was significant (greater than twice its standard error) foreach model (Anderson & Gerbing, 1988).

STUDY 1

The purpose of Study 1 was to examine the psychometric properties and thehypotheses in relation to the PPEFS by performing all of the analysesdescribed in the Analysis Strategies section above. In the following, weelaborate on the sample, procedure, and measures.

Sample and Procedure

We approached 385 employees and their managers (72) who were fromdiverse organisations and occupations. For each organisation, we identified acontact person who independently distributed our surveys to participatingemployees and managers. When the span of control of a manager exceededsix individuals, we sought to reduce manager fatigue by having the corre-sponding contact person randomly pick six of that manager’s subordinates.After the surveys were completed, each contact person retrieved the sealedsurveys from his or her organisation’s participants. The participants also hadthe option of sending their completed surveys directly back to us, theresearchers. Neither the contact person nor the participants were aware of thepurpose of the study. To avoid common method variance (CMV; Podsakoff,MacKenzie, Lee, & Podsakoff, 2003), employees filled out the survey con-taining items about the PPEFS at Time 1. A week later (Time 2), they wereasked to respond to a second survey regarding their job satisfaction, intent toquit, OCB, and demographic characteristics. At the same time (Time 2), theemployees’ immediate managers were asked to evaluate their own demo-graphic characteristics and the employees’ in-role behavior. All participants

2 We sincerely thank the Editor for suggesting that we estimate whether our PPEFS measureswere superordinate or aggregate. This investigation allowed us to join the current discussionsregarding the function and dimensionality of PE fit.

12 CHUANG ET AL.




were assured of the confidentiality of their responses. The response rate was89 per cent (343) for the employee survey and 93 per cent (67) for the managersurvey. The final usable matched data comprised 328 dyads. Of the employeeparticipants, 37 per cent were male, the average age was 30.97 years old, andthe mean organisational tenure was 55.50 months. Of the participating man-agers, 40 per cent were male, the average age was 38.45 years old, and themean organisational tenure was 104.60 months.

ε1

ε2

ε3

ε4

ε6

ε7

ε5

ε8

ε9

ε10

ε11

ε12

ε13

ε14

ε15

ε16

ε17

ε18

ε19

ε20

ε21

ε22

ε23

ε24

ε25

ε26

Item2

Item3

Item1

Item4

Item5

Item7

Item8

Item6

Item9

Item10

Item12

Item13

Item11

Item14

Item15

Item17

Item18

Item16

Item19

Item20

Item22

Item23

Item21

Item24

Item25

Item26

Values

Goals

Values

Goals

Attributes

PSFS

PJFS

POFS

PGFS

PPEFS

OCB

In-role behavior

Job satisfaction

Intent to quit

ε27

ε28

ε29

ε30

ε32

ε33

ε31

ε34

ε35

ε36

ε37

ε38

ε39

ε40

ε41

ε42

ε43

ε44

.……

……

...

ε58

ε59

ε60

ε61

ε62

Item28

Item29

Item27

Item30

Item31

Item33

Item34

Item32

Item35

Item36

Item38

Item39

Item37

Item40

Item41

Item43

Item44

Item42

……

……

…..

Item58

Item59

Item60

Item61

Item62

FIGURE 1. A model of superordinate PPEFS and outcome variables.

Note: PPEFS = Perceived Person–Environment Fit Scale; PJFS = Person–Job FitScale; POFS = Person–Organisation Fit Scale; PGFS = Person–Group Fit Scale;PSFS = Person–Supervisor Fit Scale; OCB = Organisational CitizenshipBehavior. The aggregate model was identical to the superordinate modelexcept the arrows pointed from the four fit measures to PPEFS.



78 CHUANG ET AL.


Measures

Existing Fit Measures to Test Convergent Validity and Usefulness. ForPJ fit, we used the three-item DA fit measure and the three-item NS fitmeasure from Cable and DeRue (2002). Sample items include “My abilitiesand training are a good fit with the requirements of my job” for DA fit, and“The attributes that I look for in a job are fulfilled very well by my presentjob” for NS fit. Reliability was .90 for DA fit and .91 for NS fit. We alsoadopted Cable and DeRue’s three-item PO fit measure. A sample item is “Myorganisation’s values and culture provide a good fit with the things that Ivalue in life.” Reliability was .94. We adjusted Cable and DeRue’s three-itemPO fit measure to assess person–group (PG) fit, an approach consistent withGreguras and Diefendorff (2009) and Seong and Kristof-Brown (2012). Asample item is “My personal values match my group’s values and culture.”Reliability was .92. Finally, we re-worded Cable and DeRue’s three-item POfit measure to capture PS fit, following the approach taken by Hoffman et al.(2011). A sample item is “The things that I value in life are very similar to thethings that my supervisor values.” Reliability was .92.

Measures to Test the Discriminant Validity. These included employeeage (in years) and gender (male = 1, female = 0). This practice followed thestrategy used by Liao and Chuang (2004).

Outcome Variables to Test Criterion-Related Validity and Usefulness, as

well as the Contributions and Function of the PE Fit Measures. These con-sisted of manager-evaluated employee in-role behavior, and employee-reported job satisfaction, intent to quit, and OCB. For employee in-rolebehavior, we used the four items from Van Dyne and LePine (1998), a sampleof which is “This particular employee performs the tasks that are expected aspart of the job.” Reliability was .92. For job satisfaction, we adopted the fiveitems from Brayfield and Rothe (1951), a sample of which is “I feel fairlysatisfied with my present job.” Reliability was .82. For intent to quit, we usedthe three items from Lauver and Kristof-Brown (2001), a sample of which is“If I have my way, I won’t be working for this company a year from now.”Reliability was .83. For OCB, we used the 24 items from Podsakoff,MacKenzie, Moorman, and Fetter (1990), a sample of which is “I willinglyhelp others who have work-related problems.” Reliability was .92.

STUDY 2

We conducted a second study to replicate the analyses regarding discriminantvalidity and criterion-related validity. The sample for Study 2 differed fromthat of Study 1 in that it contained employees and managers working inservice jobs, facilitating the generalisability of the PPEFS.

14 CHUANG ET AL.




Sample and Procedure

A total of 222 service-type employees and 55 managers across a variety ofindustries, organisations, and occupations participated in this study. At eachorganisation, we identified a contact person who administered one set ofsurveys to the employees and a separate set of surveys to the managers. Thesame study procedures applied in Study 1 (e.g. retrieval of the completedsurveys, assurance of confidentiality, etc.) were applied in Study 2. Employ-ees responded to scales regarding the PPEFS, job satisfaction, and personaldemographic characteristics. Supervisors provided their own demographiccharacteristics and rated the employees’ job performance. After responseswith missing data were discarded, the final sample comprised 204 employeesand 55 managers. Of the employees, 36 per cent were male, the average agewas 30.00 years old, and the mean organisational tenure was 31.60 months.Of the managers, 59 per cent were male, the average age was 36.40 years old,and the mean organisational tenure was 55.20 months.

Measures

Measures to Test Discriminant Validity. We followed Liao andChuang’s (2004) practice of adopting demographics to test discriminantvalidity. Our measures included employee age (in years) and gender (male =1, female = 0).

Measures to Test Criterion-Related Validity. These were employee jobsatisfaction and job performance. For job satisfaction, we used the five itemsfrom Brayfield and Rothe (1951), a sample of which is “I find real enjoymentin my work.” Reliability was .82. For job performance, we used the seven jobperformance items from Liao and Chuang (2004), a sample of which is“Asking good questions and listening to find out what a customer wants.”Reliability was .93.

OVERALL RESULTS

This section presents the results from both Study 1 and Study 2. Table 1 andTable 2 show the descriptive statistics and intercorrelations among the studyvariables in Study 1 and Study 2, respectively. Table 3 shows the summary ofthe results by analysis approaches across the two studies.

Stage 1: CFA and Reliability

The results of CFA are presented in Table 4. For PJFS, we selected theone-factor model (Model 1; χ2 = 3.23, df = 2, p < .20; χ2/df = 1.62; RMSEA =



80 CHUANG ET AL.


TAB

LE1

Mea

ns,

Sta

nd

ard

Dev

iati

on

s,R

elia

bili

ties

,an

dIn

terc

orr

elat

ion

sam

on

gS

tud

yVa

riab

les

inS

tud

y1

Vari

able

Mea

nSD

12

34

56

78

910

11

12

13

14

15

16

1.G

ende

r(M

gr)

.40

.49

2.A

ge(M

gr)

38.4

57.

57−.

213.

Gen

der

(Ee)

.37

.48

.30*

−.26

*4.

Age

(Ee)

30.9

77.

10−.

06.3

8**

−.04

5.P

JFS

4.59

.84

.30*

−.18

.08

.14*

*(.

84)

6.P

OF

S4.

31.8

9.3

9**

−.19

.14*

.09

.38*

*(.

91)

7.P

GF

S4.

52.6

9.2

9*−.

20.1

2*−.

03.5

3**

.68*

*(.

89)

8.P

SFS

3.97

.98

.33*

*−.

35**

.06

.08

.32*

*.6

0**

.56*

*(.

90)

9.C

&D

-PJ

fit4.

44.9

1.3

1*−.

06.0

7.1

9**

.79*

*.4

0**

.49*

*.3

2**

(.88

)10

.C

&D

-PO

fit4.

15.9

4.2

6*−.

10.1

0.1

4*.3

5**

.76*

*.5

7**

.59*

*.3

9**

(.94

)11

.C

&D

-Ada

pted

PG

fit4.

43.8

2.1

8−.

19.0

6.0

7.4

9**

.44*

*.6

8**

.44*

*.4

5**

.46*

*(.

92)

12.

C&

D-A

dapt

edP

Sfit

4.20

.98

.37*

*−.

17.0

2.1

5**

.36*

*.5

5**

.52*

*.7

5**

.36*

*.5

5**

.44*

*(.

92)

13.

In-r

ole

beha

vior

5.03

.82

.15

−.27

*.1

1*.0

2.4

1**

.25*

*.4

2**

.16*

*.2

8**

.17*

*.3

8**

.21*

*(.

92)

14.

Job

sati

sfac

tion

4.39

.84

.21

−.16

.12*

.12*

.57*

*.5

1**

.57*

*.4

1**

.56*

*.4

8**

.56*

*.4

3**

.34*

*(.

82)

15.

Inte

ntto

quit

3.94

1.16

−.26

*.0

1−.

15**

−.19

**−.

36**

−.42

**−.

35**

−.32

**−.

43**

−.37

**−.

26**

−.29

**−.

11*

−.59

**(.

83)

16.

OC

B4.

81.5

9.2

0−.

18.1

6**

.15*

*.4

4**

.43*

*.5

1**

.31*

*.3

4**

.33*

*.4

1**

.32*

*.6

0**

.44*

*−.

26**

(.92

)

No

te:

Mgr

=M

anag

er;E

e=

Em

ploy

ee;P

JFS

=P

erso

n–Jo

bF

itSc

ale;

PO

FS

=P

erso

n–O

rgan

isat

ion

Fit

Scal

e;P

GF

S=

Per

son–

Gro

upF

itSc

ale;

PSF

S=

Per

son–

Supe

rvis

orF

itSc

ale;

C&

D=

Cab

lean

dD

eRue

(200

2);

PJ

fit=

Per

son–

Job

fit;

PG

fit=

Per

son–

Gro

upfit

;P

Ofit

=P

erso

n–O

rgan

isat

ion

fit;

PS

fit=

Per

son–

Supe

rvis

orfit

;O

CB

=O

rgan

isat

iona

lCit

izen

ship

Beh

avio

r.*

p<

.05;

**p

<.0

1.

16 CHUANG ET AL.




TAB

LE2

Mea

ns,

Sta

nd

ard

Dev

iati

on

s,R

elia

bili

ties

,an

dIn

terc

orr

elat

ion

sam

on

gS

tud

yVa

riab

les

inS

tud

y2

Va

ria

ble

Mea

nSD

12

34

56

78

91

0

1.G

ende

r(M

gr)

.59

.50

2.A

ge(M

gr)

36.4

08.

53.1

23.

Gen

der

(Ee)

.36

.48

.29*

*.0

34.

Age

(Ee)

30.0

08.

63.2

2**

.35*

*.2

1**

5.P

JFS

4.78

.95

.05

.11

.08

.18*

(.89

)6.

PO

FS

6.85

1.42

.20*

*−.

02.0

8.1

2.6

0**

(.93

)7.

PG

FS

4.51

.81

.28*

*−.

04.1

5.0

4.6

1**

.86*

*(.

91)

8.P

SFS

4.04

1.13

.22*

*−.

14*

.11

.02

.36*

*.5

7**

.62*

*(.

91)

9.Jo

bsa

tisf

acti

on4.

73.9

0.0

7−.

05−.

03.0

6.6

0**

.66*

*.6

4**

.49*

*(.

82)

10.

Job

perf

orm

ance

8.49

1.75

.17*

−.29

**.1

3−.

05.2

4**

.20*

*.2

5**

.35*

*.2

0**

(.93

)

No

te:

Mgr

=M

anag

er;E

e=

Em

ploy

ee;P

JFS

=P

erso

n–Jo

bF

itSc

ale;

PO

FS

=P

erso

n–O

rgan

isat

ion

Fit

Scal

e;P

GF

S=

Per

son–

Gro

upF

itSc

ale;

PSF

S=

Per

son–

Supe

rvis

orF

itSc

ale.

*p

<.0

5;**

p<

.01.



82 CHUANG ET AL.


TAB

LE3

Su

mm

ary

of

Res

ult

sb

yth

eP

PE

FSM

easu

res

and

An

alys

isS

tag

eac

ross

Stu

die

s

An

aly

sis

Sta

ge

Fit

Mea

sure

PJ

FS

PO

FS

PG

FS

PS

FS

Subs

cale

nam

eP

JFS

PO

FS-

Val

ues

PO

FS-

Goa

lsP

GF

S-V

alue

sP

GF

S-G

oals

PG

FS-

Att

ribu

tes

PSF

S

Con

tent

dim

ensi

onK

SAs,

pers

onal

ity,

inte

rest

s,an

djo

bch

arac

teri

stic

sV

alue

san

dgo

als

Val

ues,

goal

s,an

dat

trib

utes

Val

ues,

pers

onal

ity,

wor

kst

yle,

lifes

tyle

,and

lead

ersh

ipst

yle

Num

ber

ofit

ems

4V

alue

s:4

Goa

ls:3

Val

ues:

4G

oals

:3A

ttri

bute

s:3

5

Ana

lyse

sto

test

fitm

easu

res

indi

vidu

ally

Rel

iabi

lity

.84

.91

.89

.90

Con

verg

ent

valid

ity

Supp

orte

dSu

ppor

ted

Supp

orte

dSu

ppor

ted

Dis

crim

inan

tva

lidit

yL

arge

lysu

ppor

ted

Lar

gely

supp

orte

dL

arge

lysu

ppor

ted

Supp

orte

dC

rite

rion

-rel

ated

valid

ity

Supp

orte

dSu

ppor

ted

Supp

orte

dSu

ppor

ted

Use

fuln

ess

Supp

orte

dSu

ppor

ted

Supp

orte

dSu

ppor

ted

Ana

lyse

sto

test

fitm

easu

res

sim

ulta

neou

sly

Rel

atio

nsam

ong

fitm

easu

res

Rel

ated

but

dist

inct

Rel

ativ

eim

port

ance

ofth

efit

mea

sure

sD

iffe

rent

fitm

easu

res

expl

aine

dth

egr

eate

stam

ount

ofva

rian

cein

diff

eren

tou

tcom

es

Fun

ctio

nof

the

fitm

easu

res

Supe

rord

inat

e

No

te:

PP

EF

S=

Per

ceiv

edP

erso

n–E

nvir

onm

ent

Fit

Scal

e;P

JFS

=P

erso

n–Jo

bF

itSc

ale;

PO

FS

=P

erso

n–O

rgan

isat

ion

Fit

Scal

e;P

GF

S=

Per

son–

Gro

upF

itSc

ale;

PSF

S=

Per

son–

Supe

rvis

orF

itSc

ale;

KSA

s=

Kno

wle

dge,

Skill

s,an

dA

bilit

ies.

18 CHUANG ET AL.




TAB

LE4

Res

ult

so

fC

on

firm

ato

ryFa

cto

rA

nal

ysis

,C

on

verg

ent

and

Dis

crim

inan

tVa

lidit

y,an

dFu

nct

ion

of

PP

EFS

Mo

del

Des

crip

tio

nχ2

dfχ2 /d

fR

MS

EA

CF

IIF

I

Con

firm

ator

yF

acto

rA

naly

sis

Mod

el1

PJF

Son

e-fa

ctor

mod

el3.

232

1.62

0.04

1.00

1.00

Mod

el2

PJF

Stw

o-fa

ctor

mod

ela

2.26

12.

260.

061.

001.

00M

odel

3P

OF

Sse

cond

-ord

erm

odel

b30

.83*

*13

2.37

0.07

0.99

0.99

Mod

el4

PO

FS

one-

fact

orm

odel

323.

3114

23.0

90.

260.

810.

81M

odel

5P

GF

Sse

cond

-ord

erm

odel

c87

.52*

*32

2.74

0.07

0.98

0.98

Mod

el6

PG

FS

one-

fact

orm

odel

590.

7535

16.8

80.

220.

850.

85M

odel

7P

SFS

one-

fact

orm

odel

6.50

51.

300.

031.

001.

00

Con

verg

ent

and

Dis

crim

inan

tV

alid

ity

Mod

el8

Seco

nd-o

rder

mod

eld

1220

.86

288

4.24

0.10

0.94

0.94

Mod

el9

Fou

r-fa

ctor

mod

ele

1450

.02

293

4.95

0.11

0.94

0.94

Mod

el10

One

-fac

tor

mod

el32

47.6

529

910

.86

0.17

0.87

0.87

Fun

ctio

nof

PP

EF

SM

odel

11Su

pero

rdin

ate

mod

el64

55.7

3**

1816

3.55

0.09

0.92

0.92

Mod

el12

Agg

rega

tem

odel

9332

.83

1810

5.16

0.11

0.89

0.89

Mod

el13

Dis

tinc

tm

odel

8992

.90

1802

4.99

0.11

0.89

0.89

No

te:

n=

328.

PP

EF

S=

Per

ceiv

edP

erso

n–E

nvir

onm

entF

itSc

ale;

RM

SEA

=R

ootM

ean

Squa

reE

rror

ofA

ppro

xim

atio

n;C

FI

=C

ompa

rati

veF

itIn

dex;

IFI

=In

crem

enta

lF

itIn

dex;

PJF

S=

Per

son–

Job

Fit

Scal

e;P

OF

S=

Per

son–

Org

anis

atio

nF

itSc

ale;

PG

FS

=P

erso

n–G

roup

Fit

Scal

e;P

SFS

=P

erso

n–Su

perv

isor

Fit

Scal

e.a

The

two

fact

ors

cont

aine

dD

Afit

and

NS

fit.b

The

two

subs

cale

s(P

OF

S-V

alue

san

dP

OF

S-G

oals

)wer

esp

ecifi

edas

man

ifes

tati

ons

ofa

mor

ege

nera

lum

brel

laco

nstr

uct:

PO

fit.c

The

thre

esu

bsca

les

(PG

FS-

Val

ues,

PG

FS-

Goa

ls,a

ndP

GF

S-A

ttri

bute

s)w

ere

spec

ified

asm

anif

esta

tion

sof

am

ore

gene

ralu

mbr

ella

cons

truc

t:P

Gfit

.dT

hese

cond

orde

rco

ntai

ned

the

four

PP

EF

Sm

easu

res

(i.e

.PJF

S,P

OF

S,P

GF

S,an

dP

SFS)

and

the

first

orde

rin

clud

edtw

osu

bsca

les

ofP

OF

San

dth

ree

subs

cale

sof

PG

FS.

eR

emov

edth

efir

st-o

rder

subs

cale

s.**

p<

.01.



84 CHUANG ET AL.


.04; CFI = 1.00; IFI = 1.00) because Model 2 did not result in a significantlybetter fit than Model 1 (Δχ2 = .97, Δdf = 1). Further, we picked POFS withtwo sub-scales because Model 3 was significantly better than Model 4 (Δχ2 =292.48, Δdf = 1). Our PGFS showed three sub-scales because Model 5 wassignificantly better than Model 6 (Δχ2 = 503.23, Δdf = 3). Finally, for thePSFS, the hypothesised Model 7 was acceptable. Because there were notheory- or method-based rival models to be formed, no chi-square differencetest was performed. For all of the four fit constructs, all factor loadings weresignificant. Reliability estimates were .84 for PJFS, .91 for POFS, .89 forPGFS, and .90 for PSFS.

Stage 2: Convergent, Discriminant, andCriterion-Related Validity

Convergent Validity. We expected an association between each of ourPPEFS measures and the corresponding existing measures we adopted.Results show that the correlation was .79 (p < .01) for PJFS, .76 (p < .01) forPOFS, .68 (p < .01) for PGFS, and .75 (p < .01) for PSFS. The above evidenceshows that all of the four measures exhibited acceptable convergent validity.

Discriminant Validity. We used age and gender demographics to testdiscriminant validity. Results from Study 1 show that the PJFS was notsignificantly correlated with gender (r = .08, ns), but was significantly corre-lated with age (r = .14, p < .01), providing partial support for the discriminantvalidity of the PJFS. Results show that the POFS did not have a significantcorrelation with age (r = .09, ns), although it did with gender (r = .14, p < .05),thereby suggesting partial support for the discriminant validity of the POFS.The PGFS was not significantly correlated with age (r = −.03, ns), but waswith gender (r = .12, p < .05), also implying partial support for its discrimi-nant validity. The PSFS was not correlated with either employee gender (r =.06, ns) or age (r = .08, ns), providing strong support for its discriminantvalidity. Results from Study 2 show that none of the PPEFS measures wasrelated to employee gender or age, with the exception of the PJFS, which wassignificantly related to employee age (r = .18, p < .05). Thus, discriminantvalidity was obtained for the PPEFS.

Criterion-Related Validity. Results from Study 1 reveal that all fourmeasures of the PPEFS were significantly (all ps < .01) correlated with in-rolebehavior (rs = .16–.42), job satisfaction (rs = .41–.57), intent to quit (rs =−.32–.42), and OCB (rs = .31–.51). Therefore, criterion-related validity wasobtained. Study 2 shows that all PPEFS measures were significantly (all ps <.01) related to employee job satisfaction (rs = .49–.66) and employee jobperformance (rs = .20–.35). Thus, criterion-related validity was replicated.

20 CHUANG ET AL.




Stage 3: Usefulness Analysis

For PJFS, with procedure 1 (adding Cable and DeRue’s PJ fit measure in thefirst step and then the PJFS in the second step), the addition of the PJFSsignificantly increased the multiple correlations. With procedure 2 (adding thePJFS in the first step and then the existing PJ fit measure in the second step),the existing job fit measure predicted the criteria over and above the PJFS inonly two of the four cases. For POFS, with procedure 1, changes in multiplecorrelations were significant in all four cases. With procedure 2, Cable andDeRue’s measure had additional contributions in only two of the four cases.For PGFS, with procedure 1, the changes in multiple correlations weresignificant for all four outcome variables. With procedure 2, the PGFS hadincremental validity for only two outcome variables. For PSFS, with proce-dure 1, except for one case, the addition of the PSFS significantly increased themultiple correlations. The same pattern of results was present for procedure 2.This finding indicates that the PSFS and the existing scale both contributedsomething unique (Judge et al., 2003). The above results show that the PPEFSmeasures demonstrated incremental validity and made unique contributionsabove and beyond the measures’ respective existing measure.

Stage 4: Relations among the PPEFS Measures, andConvergent and Discriminant Validity of thePPEFS Measures

Findings from the preceding validations enabled us to identify four fit meas-ures (i.e. PJFS, POFS, PGFS, and PSFS). Pearson correlations among thefour measures revealed that all relationships were significant (rs = .32–.68, allps < .01; mean r = .51). The convergent validity of the measures was supportedby the existence of a reasonable fit of the proposed second-order model (seeTable 4, Model 8) and by the fact that all items loaded significantly on theirrespective construct. Discriminant validity was also achieved, as none of theconfidence intervals around the correlation estimate between the two factorsincluded 1.0. Finally, chi-square difference tests revealed that the second-ordermodel fit significantly better than the four-factor model (Δχ2 = 229.16, Δdf = 5;see Table 4, Model 9) and the one-factor model (Δχ2 = 2026.79, Δdf = 11; seeTable 4, Model 10). Taken together, these results imply that while the presentstudy’s different fit perceptions might correlate to one another, they aredistinguishable and can serve as distinct constructs in future studies.

Stage 5: Contributions of the PPEFS Measures

We predicted that PJ fit (Hypotheses 1a–1d), PO fit (Hypotheses 2a–2d), PGfit (Hypotheses 3a–3d), and PS fit (Hypotheses 4a–4d) are each related to



86 CHUANG ET AL.


in-role behavior, job satisfaction, intent to quit, and OCB. Results shown inTable 5 indicate that all hypotheses were supported except for Hypotheses3c, 4a, and 4c. In terms of relative weight, results show that different fitmeasures explained the greatest amount of variance in different outcomes.For example, the greatest amount of variance of in-role behavior wasexplained by PGFS (45%), followed by PJFS (40%), POFS (10%), and PSFS(5%).

Stage 6: Function (Superordinate vs. Aggregate) of thePPEFS Measures

Hypothesis 5a postulated that PE fit is an aggregate multidimensional con-struct, whereas Hypothesis 5b postulated that PE fit is a superordinate mul-tidimensional construct. Results of the superordinate model show anacceptable model fit (see Table 4, Model 11). All factor loadings and paths ofthe model were significant. The results of the aggregate model show anundesirable model fit (see Table 4, Model 12). None of the paths from thefour fit measures to the PPEFS or from the PPEFS to the outcomes weresignificant. Therefore, Hypothesis 5a was not supported, but Hypothesis 5bwas supported. In a supplemental analysis, we estimated a distinct model,where we removed the PPEFS to estimate direct relationships between the

TABLE 5Results of Relative Weight Analysis (RWA) of the PPEFS on

Outcomes Variablesa

In-role behaviorb Job satisfaction Intent to quit OCB

RW %RW RW %RW RW %RW RW %RW

PJFS .08** 40% .18** 41% .06** 30% .06** 24%POFS .02* 10% .09** 20% .07** 35% .06** 24%PGFS .09** 45% .12** 27% .03 15% .11** 44%PSFS .01 5% .05** 12% .04 20% .02* 8%Total R2c .20 .44 .20 .25

Note: PPEFS = Perceived Person–Environment Fit Scale; PJFS = Person–Job Fit Scale; POFS = Person–Organisation Fit Scale; PGFS = Person–Group Fit Scale; PSFS = Person–Supervisor Fit Scale; OCB =Organisational Citizenship Behavior; RW = Relative Weights in R2 Form (Johnson, 2000); %RW = RelativeWeights in Percentage Form (calculated by dividing individual relative weights by the total R2 and multiplyingby 100). a We partialled out the effects of control variables from each predictor (i.e. fit measure) following thestrategy recommended in LeBreton, Tonidandel, and Krasikova (2013). To remove the effects of controlvariables from outcomes, we then regressed each outcome on the residualised predictors and the controlvariables. Control variables are employees’ age, gender, education, and tenure. b Manager-reported measure.c R2 summed across the four fit measures.* p < .05; ** p < .01.

22 CHUANG ET AL.




four fit measures and the outcomes with all possible combinations. Theresults also show an undesirable model fit (see Table 4, Model 13). Most ofthe paths from the four measures to the outcomes were either non-significantor had an unexpected sign. In conclusion, the superordinate model betterrepresented the relationships of the four fit measures.

Finally, we address the concern of CMV by following the procedural andstatistical remedies recommended by Podsakoff et al. (2003). Procedurally,we assured our participants that their responses would be confidential. Inaddition, we collected data from two sources (i.e. employees and supervisors)at two different points in time. Statistically, we tested Harman’s single-factormodel using principal component factor analysis with an unrotated solutionand found that the first factor accounted for only 28 per cent of the variance.Second, we estimated and compared a series of measurement models includ-ing a one-factor model, a two-factor model (by source), another two-factormodel (by time), and a hypothesised eight-factor model (i.e. PJ fit, PO fit, PGfit, PS fit, in-role behavior, job satisfaction, intent to quit, and OCB). Resultsrevealed that the model fit of the eight-factor model was acceptable (χ2 =7092.84, df = 1823, p < .01; χ2/df = 3.89; RMSEA = .09; CFI = .91; IFI = .91)and was significantly better than the other three models. Lastly, followingPodsakoff et al., we estimated the superordinate model (Figure 1) with anunmeasured latent CMV construct as compared with one without the CMVconstruct. Results indicated that the model with CMV inevitably generatedan improved fit, but none of the factor loadings of the CMV factor weresignificant. In addition, all path coefficients in these two models were signifi-cant, and the corresponding coefficient magnitudes between these twomodels were highly similar. All of the above results suggest that the relation-ships among the constructs are not largely affected by common method bias.

GENERAL DISCUSSION

Our study is among the first to develop a comprehensive perceived person–environment fit scale based on multiple theories. We conducted two inde-pendent studies that involved both multiple rater sources and data collectionat multiple points in time. Our results suggested four measures of PE fit (i.e.PJ fit scale, PO fit scale, PG fit scale, and PS fit scale). When examinedindividually, these measures’ reliability values are all well above the criticallevel of .70 for newly developed scales (Nunnally, 1976). All measures dem-onstrate convergent validity with relevant existing measures, discriminantvalidity with demographic variables, and criterion-related validity with aseries of theory-related criterion variables. Furthermore, the usefulness of themeasures is obtained since the measures show incremental validity above andbeyond an existing corresponding fit measure in the majority of cases. Wetested the four measures simultaneously as well. The results of correlational



88 CHUANG ET AL.


analysis and discriminant validity testing of the measurement model suggestthat the four measures are related but distinct. Besides, the relative impor-tance of each measure differs largely by outcome variables. Lastly, the fourmeasures are reflective of a superordinate construct of PE fit. Thus, this studyoffers sound empirical evidence of these new scales of PE fit perceptions.

Theoretical Contributions and Implications

This research into developing multidimensional scales for fit perceptions hasseveral theoretical implications. First, our study constitutes a response to atwo-decade-old call for research on the dimensionality of fit measures(Edwards & Cooper, 1990; Piasentin & Chapman, 2006). We respondedto the call by incorporating multiple dimensions of PE fit into a newscale. Recent PE fit research has started to examine fit along the linesof a multidimensional theory (Edwards & Billsberry, 2010; Jansen &Kristof-Brown, 2006; Wheeler et al., 2005). For instance, Jansen and Kristof-Brown proposed a nested view for incorporating person–vocation, person–organisation, person–group, person–job, and person–person fit into anintegrative multidimensional theory of PE fit. Edwards and Billsberry’stesting of the multidimensional PE fit model advanced by Jansen andKristof-Brown appears to be the first empirical test of multiple dimensions ofPE fit. These authors found that the different forms of fit separately affectedthe outcomes of employee commitment, intent to quit, and job satisfaction.These recent advances in PE fit research have significantly contributed to aholistic and realistic view of how people experience PE fit.

Nevertheless, despite having recognised the importance of viewing PE fitfrom a multidimensional perspective, the field still needs a multidimensionalPE fit instrument that is guided by an integration of the existing PE fittheories. Fit studies have traditionally captured the many dimensions of PEfit by gathering measures from different sources (i.e. studies) with varyingmethods or formats. The drawback of this tradition is that the effects of thevarious types of PE fit vary not only because of the true variance of the fitconstruct but also because of the distinct methods. Our scale, the PPEFS,integrates multiple fit-related theories to provide a psychometrically soundtool that incorporates a full spectrum of fit dimensions.

Second, our study addresses the relations and differentiation among fourdimensions of fit perceptions. Correlation analysis shows that the differenttypes of PE fit are related to one another. Although this finding is largelyconsistent with Kristof-Brown et al.’s (2005) meta-analytical findings, onesalient difference is found for the correlation between PJ fit and PO fit.Specifically, the meta-analytical result shows a correlation of .74 (p < .05) fordirect measures of PJ fit and PO fit, whereas our result shows a correlation of.38 (p < .01). We surmise that one reason behind this large gap might be the

24 CHUANG ET AL.




lack of specifications of content dimensions in some studies.3 In comparison,Lauver and Kristof-Brown (2001) measured PJ fit using specific contentdimensions such as skills, abilities, and personality, and measured PO fitusing a content dimension of values, much like the content dimensions usedin our study. They found a correlation of .18 (p < .05). Other studies usingspecific content dimensions resulted in lower correlations as well (e.g.Erdogan & Bauer, 2005). This preliminary evidence seems to underscore theimportance of an unambiguous dimensionality of PE fit via the developmentof a multidimensional PE fit scale where the dimensionality of PE fit is welldefined by theory.

Third, our research reveals that, instead of being an aggregate or a distinctconstruct, PE fit is better conceptualised as a superordinate multidimensionalconstruct manifested by its dimensions. Our result is similar to the findings ofSeong and Kristof-Brown (2012) and Seong et al. (in press) which show that“the theoretical orthogonality in the [fit] concepts is not found in practice”(Seong et al., in press). Specifically, on one hand, researchers have theorisedthat people are able to distinguish between different types of fit, and thatdifferent types of fit can predict different outcomes. This implies an aggregatemodel. On the other hand, the meta analytic results found by Kristof-Brownet al. (2005) revealed that the correlations between multiple types of fit (i.e.PJ, PO, PG, and PS fit) range from .37 (p < .05) to .74 (p < .05). While themoderate to high correlations could imply either a superordinate or anaggregate model, Seong and Kristof-Brown, Seong et al., and our study allfound PE fit to be a superordinate multidimensional construct.4 Because thefield has just begun to investigate PE fit as a multidimensional construct, weencourage future research to use multidimensional fit scales to continue onthis imperative line of research.

Fourth, in order to deal with the issue of multicollinearity among thevarious fit dimensions, our study adopted relative weight analysis (Johnson,2000) to specifically consider the correlations among the dimensions while

3 For example, Kristof-Brown (2000) asked participants to measure PO fit using such generalitems as “To what degree does this applicant fit with your organisation?” and “To what extentwill other employees think this candidate fits well in your organisation?” One sample item of PJfit is “To what extent will other employees think this candidate is qualified to do this job?” Thecorrelation between PJ and PO fit was .72 (p < .05). Similarly, Saks and Ashforth (1997) utilisedsuch general items as “To what extent does your new organisation measure up to the kind oforganisation you were seeking?” (PJ fit) and “To what extent does your new job measure up tothe kind of job you were seeking?” (PO fit). The correlation was .56 (p < .01).

4 MacKenzie, Podsakoff, and Jarvis (2005) suggested that one understand theconceptualisation of multidimensional constructs from how strongly the measures are correlatedwith each other. These authors stated that a superordinate model predicts that the measuresshould be correlated with each other because they share a common cause, whereas an aggregatemodel makes no predictions about the correlations (i.e. they could be at any level).



90 CHUANG ET AL.


attempting to delineate the relationship between a specific fit dimension andthe outcomes. With this method, researchers avoid obtaining regression coef-ficients that are either not significant or have an unexpected sign (e.g. Cable& DeRue, 2002; Edwards & Billsberry, 2010). The field has barely started toexamine the relative importance of various types of fit, and we encouragefuture research to follow this line of investigation using multidimensional PEfit scales in order to accumulate more evidence.

Finally, our PPEFS adds to Cable and DeRue’s (2002) PJ fit and PO fitscales. We chose Cable and DeRue’s scales for comparison purposes becausetheir scales are frequently adopted and contain multiple dimensions for PE fit(Kristof-Brown & Guay, 2011). Based on the results of this comparison, thecontribution of the PPEFS can be summarised from the theoretical andempirical perspectives. On one hand, theoretically, the PPEFS is based onmany relevant PE fit theories. On the other hand, empirically, we prove thatadditional fit scales (i.e. PG fit and PS fit) are useful. In the decade that haspassed since Cable and DeRue’s prominent work appeared, the evolution ofthis field has accentuated the need for updated PE fit scales. As managementliterature continues to pay close attention to interpersonal relationships suchas subordinate–supervisor dyads and co-worker relationships (Oh et al.,2014), an updated instrument capable of including the PS fit and PG fit scalesbecomes an especially high priority. This need is also reflected by the fact thatrecent fit research has begun to adapt Cable and DeRue’s PO fit items(measuring values fit) to gauge PS fit and PG fit (e.g. Greguras & Diefendorff,2009; Hoffman et al., 2011; Seong & Kristof-Brown, 2012). This adaptationis less ideal since people trying to fit in with other people in a work setting(e.g. supervisor and co-workers) experience more aspects than are covered byvalues alone. Results of our usefulness analysis reveal that each of the PPEFSmeasures did indeed have incremental validity above and beyond Cable andDeRue’s corresponding scale. Nevertheless, we emphasise that the develop-ment of the PPEFS represents more than simply extra dimensions of PE fit ormere comparisons with existing scales, but also, in a more important sense, aPE fit instrument geared toward multidimensional theories of fit.5

Practical Implications

From an organisation’s point of view, having a means by which to identifydistinct fit perceptions can help human resource departments select appli-cants, recognise specific areas for improvement and training, identify areas ofrole conflict, and reassign and redesign jobs and job tasks. For instance, byusing the scores of the PSFS scale, managers can more easily identify the

5 We thank the editor and the two reviewers for encouraging us to address the contributionof our PPEFS beyond the scales of Cable and DeRue (2002).

26 CHUANG ET AL.




exact reason for a mismatch between a subordinate and a supervisor (e.g.mismatched personalities, work styles, or leadership styles). The managerswill then be able to take appropriate actions to increase congruence. Evenwith the same content dimension applied to different types of PE fit (e.g.personality in PSFS and personality in PGFS), the PPEFS has ample mana-gerial implications, because the meaning of the same content dimension mayvary across different fit phenomena. For example, an extroverted individualmay find him- or herself to be a good match with an extroverted supervisorand yet have a hard time “fitting in” with introverted co-workers in the sameenvironment. Managers should be trained to recognise the different types offit that employees experience to be better able to help employees with variousneeds. We suggest that organisations treat and use the four fit measuresconstructed herein as helpful managerial tools.

Limitations and Future Research

This current study has several limitations that point to possible researchdirections. First, sampling bias may be an issue for our study because weinstructed our organisational contact persons to select employee participants.In addition to the procedural accounts described in the Sample and Proce-dure section, we performed the following empirical tests. First, we conductedt-tests on data from Study 2 in order to compare the usable and unusablecases with regard to the means of employee demographics, our PPEFS meas-ures, and the outcome variables. Results indicate that the two groups werenot statistically significantly different from each other, indicating no sam-pling bias. We did not perform t-tests for Study 1 because the unusable caseswere not available. Second, a range restriction did not prevail in our researchbecause we found that a majority of the correlations between our PPEFSmeasures and the theoretically relevant outcome variables were significant.Third, the means and standard deviations of our PE fit measures were com-parable to the means and standard deviations in existing PE fit studies. Basedon this evidence, we hope that sampling bias might not represent a seriousconcern in our research.

Second, future research could extend our study to examine the conditionsin which the activation of certain dimensions is greater than the activation ofothers. Kristof-Brown and Guay (2011) acknowledged that identifying whycertain fit dimensions are salient is an important issue now that the multidi-mensionality of PE fit is recognised. This is important since having manydistinct aspects of fit leads to a greater possibility of conflict among dimen-sions. Wheeler et al. (2005) stated that individual preferences or environmen-tal cues may cause some fit dimensions to be more salient than others.Cognitive dissonance may occur when personal preferences and social cuesactivate different fit dimensions. However, having distinct dimensions of fit



92 CHUANG ET AL.


also has benefits such as allowing one to create buffers between negative andpositive fit experiences. Future research should consider studying the condi-tions in which—or the types of people for which—cognitive dissonance orbuffering effects could take place.

Conclusion

By integrating various PE fit theories together for the development of amultidimensional instrument of perceived PE fit, our two studies follow aprominent current trend of PE fit research, which centers on clarifying themultidimensionality of PE fit. The results of a series of scale developmentvalidations and hypothesis tests reveal our PPEFS to be a psychometricallysound instrument. The many fit dimensions of this new instrument are dis-tinct and proven to incrementally contribute to commonly adopted existingfit scales. With the development of this scale, we also advance the fit literatureby identifying the relative weight and function of the fit dimensions. Our newinstrument also contributes to the current theoretical and practical under-standing of PE fit. We hope that, through the breadth, depth, and rigor of ourapproach, our instrument will prove useful in guiding future research regard-ing PE fit.

REFERENCES

Anderson, J.C., & Gerbing, D.W. (1988). Structural equation modeling in practice: Areview and recommended two-step approach. Psychological Bulletin, 103, 411–423.

Bentler, P.M., & Bonett, D.G. (1980). Significance tests and goodness of fit in theanalysis of covariance structures. Psychological Bulletin, 88, 588–606.

Brayfield, A.H., & Rothe, H.F. (1951). An index of job satisfaction. Journal of

Applied Psychology, 35, 307–311.Bretz, R.D., Jr., Rynes, S.L., & Gerhart, B. (1993). Recruiter perceptions of applicant

fit: Implications for individual career preparation and job search behavior. Journal

of Vocational Behavior, 43, 310–327.Byrne, D. (1971). The attraction paradigm. New York: Academic Press.Cable, D.M., & DeRue, D.S. (2002). The convergent and discriminant validity of

subjective fit perceptions. Journal of Applied Psychology, 87, 875–884.Chuang, A., Judge, T.A., & Liaw, Y.J. (2012). Transformational leadership and

customer service: A moderated mediation model of negative affectivity and emotionregulation. European Journal of Work and Organizational Psychology, 21, 28–56.

Chuang, A., & Sackett, P.R. (2005). The perceived importance of person–job fit andperson–organization fit between and within interview stages. Social Behavior and

Personality, 33, 209–226.DiMarco, N. (1974). Supervisor–subordinate life style and interpersonal need com-

patibilities as determinants of subordinate’s attitudes toward the supervisor.Academy of Management Journal, 17, 575–578.

28 CHUANG ET AL.




DiMarco, N. (1975). Life style, work group structure, compatibility, and job satis-faction. Academy of Management Journal, 18, 313–322.

Edwards, J.R. (1991). Person–job fit: A conceptual integration, literature review, andmethodological critique. In C.L. Cooper & I.T. Robertson (Eds.), International

review of industrial and organizational psychology (pp. 283–357). New York: JohnWiley and Sons.

Edwards, J.A., & Billsberry, J. (2010). Testing a multidimensional theory of person–environment fit. Journal of Managerial Issues, 22, 476–493.

Edwards, J.R., & Cooper, C.L. (1990). The person–environment fit approach tostress: Recurring problems and some suggested solutions. Journal of

Organizational Behavior, 11, 293–307.Erdogan, B., & Bauer, T.N. (2005). Enhancing career benefits of employee proactive

personality: The role of fìt with jobs and organizations. Personnel Psychology, 58,859–891.

French, J.R.P., Jr., & Kahn, R.L. (1962). A programmatic approach to studying theindustrial environment and mental health. Journal of Social Issues, 18, 1–48.

Greguras, G.J., & Diefendorff, J.M. (2009). Different fits satisfy different needs:Linking person–environment fit to employee commitment and performance usingself-determination theory. Journal of Applied Psychology, 94, 465–477.

Harrison, D.A., Price, K.H., Gavin, J., & Florey, A. (2002). Time, teams, and taskperformance: Changing effects of surface and deep-level diversity on group func-tioning. Academy of Management Journal, 45, 1029–1045.

Hinkin, T.R. (1998). A brief tutorial on the development of measures for use in surveyquestionnaires. Organizational Research Methods, 1, 104–121.

Hoffman, B.J., Bynum, B.H., Piccolo, R.F., & Sutton, A.W. (2011). Person–organization value congruence: How transformational leaders influence workgroup effectiveness. Academy of Management Journal, 54, 779–796.

Holland, J.L. (1996). Exploring careers with a typology: What we have learned andsome new directions. American Psychologist, 51, 397–406.

Huang, X., & Iun, J. (2006). The impact of subordinate–supervisor similarity ingrowth-need strength on work outcomes: The mediating role of perceived similar-ity. Journal of Organizational Behavior, 27, 1121–1148.

Huston, T.L., & Levinger, G. (1978). Interpersonal attraction and relationships.Annual Review of Psychology, 29, 115–156.

Jansen, K.J., & Kristof-Brown, A. (2006). Toward a multidimensional theory ofperson–environment fit. Journal of Managerial Issues, 18, 193–212.

Johnson, J.W. (2000). A heuristic method for estimating the relative weight ofpredictor variables in multiple regression. Multivariate Behavioral Research, 35,1–19.

Judge, T.A., Erez, A., Bono, J.E., & Thoresen, C.J. (2003). The core self-evaluationsscale: Development of a measure. Personnel Psychology, 56, 303–331.

Kim, T.Y., Aryee, S., Loi, R., & Kim, S.P. (2013). Person–organization fit andemployee outcomes: Test of a social exchange model. International Journal of

Human Resource Management, 24, 3719–3737.Kristof, A.L. (1996). Person–organization fit: An integrative review of its conce-

ptualizations, measurement, and implications. Personnel Psychology, 49, 1–49.



94 CHUANG ET AL.


Kristof-Brown, A.L. (2000). Perceived applicant fit: Distinguishing between recruit-ers’ perceptions of person–job and person–organization fit. Personnel Psychology,53, 643–671.

Kristof-Brown, A.L., & Guay, R.P. (2011). Person–environment fit. In S. Zedeck(Ed.), American Psychological Association handbook of industrial and

organizational psychology (Vol. 3, pp. 3–50). Washington, DC: American Psycho-logical Association.

Kristof-Brown, A.L., Jansen, K.J., & Colbert, A.E. (2002). A policy-capturing studyof the simultaneous effects of fit with jobs, groups, and organizations. Journal of

Applied Psychology, 87, 985–993.Kristof-Brown, A.L., & Stevens, C.K. (2001). Goal congruence in project teams:

Does the fit between members’ personal mastery and performance goals matter?Journal of Applied Psychology, 86, 1083–1095.

Kristof-Brown, A.L., Zimmerman, R.D., & Johnson, E.C. (2005). Consequences ofindividuals’ fit at work: A meta-analysis of person–job, person–organization,person–group, and person–supervisor fit. Personnel Psychology, 58, 281–342.

Kulik, C.T., Oldham, G.R., & Hackman, J.R. (1987). Work design as an approach toperson–environment fit. Journal of Vocational Behavior, 31, 278–296.

Lauver, K.J., & Kristof-Brown, A.L. (2001). Distinguishing between employees’ per-ceptions of person–job and person–organization fit. Journal of Vocational

Behavior, 59, 454–470.LeBreton, J.M., Tonidandel, S., & Krasikova, D.V. (2013). Residualized relative

importance analysis: A technique for the comprehensive decomposition of vari-ance in higher order regression models. Organizational Research Methods, 16,449–473.

Li, C.K., & Hung, C.H. (2010). An examination of the mediating role of person–jobfit in relations between information literacy and work outcomes. Journal of Work-

place Learning, 22, 306–318.Liao, R., & Chuang, A. (2004). A multilevel investigation of factors influencing

employee service performance and customer outcomes. Academy of Management

Journal, 47, 41–58.McCulloch, M.C., & Turban, D.B. (2007). Using person–organization fit to select

employees for high-turnover jobs. International Journal of Selection and Assess-

ment, 15, 63–71.MacKenzie, S.B., Podsakoff, P.M., & Jarvis, C.B. (2005). The problem of measure-

ment model misspecification in behavioral and organizational research and somerecommended solutions. Journal of Applied Psychology, 90, 710–730.

MacKenzie, S.B., Podsakoff, P.M., & Podsakoff, N.P. (2011). Construct measure-ment and validation procedures in MIS and behavioral research: Integrating newand existing techniques. MIS Quarterly, 35, 293–334.

Muchinsky, P.M., & Monahan, C.J. (1987). What is person–environment congru-ence? Supplementary versus complementary models of fit. Journal of Vocational

Behavior, 31, 268–277.Nunnally, J.C. (1976). Psychometric theory. New York: McGraw-Hill.Oh, I.S., Guay, R.P., Kim, K., Harold, C.M., Lee, J.H., Heo, C.G., & Shin, K.H.

(2014). Fit happens globally: A meta-analytic comparison of the relationships of

30 CHUANG ET AL.




person–environment fit dimensions with work attitudes and performanceacross East Asia, Europe, and North America. Personnel Psychology, 67, 99–152.

Ostroff, C., Shin, Y., & Kinicki, A.J. (2005). Multiple perspectives of congruence:Relationships between value congruence and employee attitudes. Journal of

Organizational Behavior, 26, 591–623.Pervin, L.A. (1968). Performance and satisfaction as a function of individual–

environment fit. Psychological Bulletin, 69, 56–68.Piasentin, K.A., & Chapman, D.S. (2006). Subjective person–organization fit: Bridg-

ing the gap between conceptualization and measurement. Journal of Vocational

Behavior, 69, 202–221.Pierce, J.L., Gardner, D.G., Cummings, L.L., & Dunham, R.B. (1989). Organization-

based self-esteem: Construct definition, measurement, and validation. Academy of

Management Journal, 32, 622–648.Podsakoff, P.M., MacKenzie, S.B., Lee, J.Y., & Podsakoff, N.P. (2003). Common

method biases in behavioral research: A critical review of the literature and rec-ommended remedies. Journal of Applied Psychology, 88, 879–903.

Podsakoff, P.M., MacKenzie, S.B., Moorman, R., & Fetter, R. (1990). Transforma-tional leader behaviors and their effects of followers’ trust in leader, satisfactionand organizational citizenship behaviors. Leadership Quarterly, 4, 107–142.

Ravlin, E.C., & Meglino, B.M. (1987). Effect of values on perception and decisionmaking: A study of alternative work value measures. Journal of Applied Psychol-

ogy, 72, 666–673.Riordan, C.M. (2000). Relational demography within groups: Past developments,

contradictions, and new directions. In K.M. Rowland & G.R. Ferris (Eds.),Research in personnel and human resource management (Vol. 19, pp. 131–173).Greenwich, CT: JAI Press.

Saks, A.M., & Ashforth, B.E. (1997). A longitudinal investigation of the relationshipsbetween job information sources, applicant perceptions of fit, and work outcomes.Personnel Psychology, 50, 395–426.

Schaubroeck, J., & Lam, S.S.K. (2002). How similarity to peers and supervisorinfluences organizational advancement in different cultures. Academy of Manage-

ment Journal, 45, 1120–1136.Schneider, B. (1987). The people make the place. Personnel Psychology, 40, 437–454.Seong, J.Y., & Kristof-Brown, A.L. (2012). Testing multidimensional models of

person–group fit. Journal of Managerial Psychology, 27, 536–556.Seong, J.Y., Kristof-Brown, A.L., Park, W.W., Hong, D.S., & Shin, Y. (in press).

Person–group fit: Diversity antecedents, proximal outcomes, and performance atthe group level. Journal of Management.

Shaw, M.E. (1981). Group dynamics: The psychology of small group behavior (3rdedn.). New York: McGraw-Hill.

Tang, T.L.P., & Sutarso, T. (2013). Falling or not falling into temptation? Multiplefaces of temptation, monetary intelligence, and unethical intentions across gender.Journal of Business Ethics, 116, 529–552.

Turban, D.B., & Jones, A.P. (1988). Supervisor–subordinate similarity: Types,effects, and mechanisms. Journal of Applied Psychology, 73, 228–234.



96 CHUANG ET AL.


Vancouver, J.B., & Schmitt, N.W. (1991). An exploratory examination of person–organization fit: Organizational goal congruence. Personnel Psychology, 44, 333–352.

Van Dyne, L.V., & LePine, J.A. (1998). Helping and voice extra-role behaviors:Evidence of construct and predictive validity. Academy of Management Journal,41, 108–119.

Van Vianen, A.E.M. (2000). Person–organization fit: The match between newcomers’and recruiters’ preferences for organizational cultures. Personnel Psychology, 53,113–149.

Wang, M., Zhan, Y., McCune, E., & Truxillo, D. (2011). Understanding newcomers’adaptability and work-related outcomes: Testing the mediating roles of perceivedP–E fit variables. Personnel Psychology, 64, 163–189.

Wheeler, A.R., Buckley, M.R., Halbesleben, J.R.B., Brouer, R.L., & Ferris, G.R.(2005). “The elusive criterion of fit” revisited: Toward an integrative theory ofmultidimensional fit. Research in Personnel and Human Resources Management,24, 265–304.

32 CHUANG ET AL.




APPENDIX

Final Items of the Perceived Person–Environment FitScale (PPEFS)Person–Job Fit Scale (PJFS)1. How would you describe the match between your professional skills, knowledge, and abilities and

those required by the job?2. How would you describe the match between your personality traits (e.g. extrovert vs. introvert,

agreeable vs. disagreeable, and dependable vs. undependable) and those required by the job?3. How would you describe the match between your interests (e.g. social vs. unsocial, artistic vs.

inartistic, and conventional vs. unconventional) and those you desire for a job?4. How would you describe the match between the characteristics of your current job (e.g. autonomy,

importance, and skill variety) and those you desire for a job?Person–Organisation Fit Scale (POFS)

POFS-ValuesHow would you describe the match between your emphasis and your organisation’s emphasis on

the following values?1. honesty2. achievement3. fairness4. helping others

POFS-GoalsHow would you describe the match between your goals and your organisation’s goals on the

following dimensions?5. reward6. the amount of effort expected7. competition with other organisations

Person–Group Fit Scale (PGFS)PGFS-Values

How would you describe the match between your emphasis and your group’s emphasis on thefollowing values?

1. honesty2. achievement3. fairness4. helping others

PGFS-GoalsHow would you describe the match between your goals and your group’s goals on the following

dimensions?5. reward6. the amount of effort expected7. competition with other groups

PGFS-AttributesHow would you describe the match between you and your group members on the following

characteristics?8. personality9. work style

10. lifestylePerson–Supervisor Fit Scale (PSFS)1. How would you describe the match between the things you value in life and the things your

supervisor values?2. How would you describe the match between your personality and your supervisor’s personality?3. How would you describe the match between your work style and your supervisor’s work style?4. How would you describe the match between your lifestyle and your supervisor’s lifestyle?5. How would you describe the match between your supervisor’s leadership style and the leadership

style you desire?

Note: All items used a 7-point scale, 1 meaning “no match” and 7 meaning “complete match”.



98 CHUANG ET AL.


Documents

Development of a Multidimensional Instrument of Person ...web.ba.ntu.edu.tw/professor/contents/aichia/ChuangShenJudge2016APIR.pdf · multidimensional (Edwards & Billsberry, 2010;